Updated

- Issue fixing
- PDF Options support added. (beta)

Author: NilGems

composer.json

@@ -38,7 +38,7 @@
         "phpoffice/phpspreadsheet": "^1.23",
         "phpoffice/phpword": "^0.18",
         "laravel/framework": "^12.0",
-        "thiagoalessio/tesseract_ocr": "^2.12",
+        "thiagoalessio/tesseract_ocr": "^2.13",
         "html2text/html2text": "^4.3",
         "phpoffice/phppresentation": "^1.0"
     },

config/textract.php

@@ -34,7 +34,7 @@
          | For more details please visit - https://github.com/thiagoalessio/tesseract-ocr-for-php#executable
          |
          */
-        'executable_path' => env('TEXTRACT_OCR_EXEC_PATH', null),
+        'executable_path' => env('TEXTRACT_OCR_EXEC_PATH', 'tesseract'),
 
         /*
          | -------------------------------------------------------------------------------------------------------------

src/ExtractorService/Ocr/TesseractOcrRun.php

@@ -35,6 +35,8 @@ protected function getOcr(TesseractOcrOptions $ocrOptions = null): TesseractOCR
             $ocrOptions = new TesseractOcrOptions();
         }
         $ocr = new TesseractOCR($this->utilsService->getFilePath());
+        $ocr->psm(6);
+        $ocr->config('preserve_interword_spaces', '1');
         $ocr->withoutTempFiles();
         if ($ocrOptions) {
             foreach ($ocrOptions->toArray() as $option_key => $option_value) {

src/Services/UtilsService.php

@@ -151,9 +151,11 @@ public function getFileMimeType(): ?string
      * @param PdfOptions $options
      * @return $this
      */
-    public function setPdfOptions(PdfOptions $options): self
+    public function setPdfOptions(?PdfOptions $options): self
     {
-        $this->pdf_options = $options;
+        if (!is_null($options)) {
+            $this->pdf_options = $options;
+        }
         return $this;
     }